5
$\begingroup$

The following is from Tom Apostol's Calculus I, on page 250, exercise 42.:

If $\mathit{n}$ is a positive integer and if $\mathit{x} > 0$, show that
$$ \left( 1 + \frac{x}{n} \right)^n < e^x \text, \qquad \text{and that} \qquad e^x < \left(1 - \frac{x}{n}\right)^{-n} \quad \text{if} \quad x < n. $$

There is no solution provided in the book, I would like to ask someone to verify if mine is correct:

Integrating the inequality $ 1 > \frac{1}{1 + \frac{x}{n}} $ we get that

$$ \int_0^x{1dt} > \int_0^x{\frac{1}{1 + \frac{t}{n}}dt} \qquad \text{yielding} \qquad x > n\log\left(1 + \frac{x}{n}\right). $$

Since $ e $ is strictly increasing it follows that $$ e^x > \left(1 + \frac{x}{n}\right)^{n}. $$

Similarly, integrating $ 1 < \frac{1}{1 - \frac{x}{n}} $ we get that $$ \int_0^x{1dt} < \int_0^x{\frac{1}{1 - \frac{t}{n}}dt} \qquad \text{yielding} \qquad x < -n\log\left(1 - \frac{x}{n}\right) . $$ It follows that $$ e^x <\left(1 - \frac{x}{n}\right)^{-n} . $$

$\endgroup$
  • 1
    $\begingroup$ Simpler is to note that $e^x\ge1+x$ for all $x\in\mathbb{R}$. However, both your derivations look fine. $\endgroup$ – robjohn Oct 5 '15 at 20:14
  • $\begingroup$ Thanks for going through it. Could you explain how we get from your inequality to the ones in the question? $\endgroup$ – Imre Deák Oct 5 '15 at 20:24
  • $\begingroup$ I have added an answer with some more detail. $\endgroup$ – robjohn Oct 5 '15 at 20:29
  • 1
    $\begingroup$ If you look at Exercise 41 in Apostol, you'll see it asks you to deduce the inequality robjohn cites. $\endgroup$ – Barry Cipra Oct 5 '15 at 20:53
  • 1
    $\begingroup$ @ImreDeák, it's always good to step back and look for what you might have overlooked, but actually I find a lot to like in the approach you took to proving the inequalities. $\endgroup$ – Barry Cipra Oct 7 '15 at 20:22
4
$\begingroup$

Your proofs look fine.

Another approach is to use the fact that $e^x\ge1+x$ for all $x\in\mathbb{R}$, with strict inequality for $x\ne0$. This follows from the strict convexity of $e^x-1-x$ and its minimum of $0$ at $x=0$.

Then we have for $x\gt0$, $$ e^{x/n}\gt1+\frac xn\implies e^x\gt\left(1+\frac xn\right)^n\tag{1} $$ and $$ e^{-x/n}\gt1-\frac xn\implies e^x\lt\left(1-\frac xn\right)^{-n}\tag{2} $$ Inequality $(2)$ does rely on the fact that $x\lt n$ so that the quantity being raised to the negative power is positive.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.